Intelligent Hybrid Multi-Agent Computational Architecture for Resources Management in Complex Systems
G.A. Aderounmu, F.J. Ogwu, E.R. Adagunodo, A.A. Akintola, A.O. Adetoye, P.K. Mahanti
An Intelligent hybrid multi-agent architecture is hereby presented for system resource management. The architecture was based on the concepts of systems decomposition and synthetically integration of subsystems. The architecture was defined both at the task structure and the computational levels, and was made up of five processing phases. These phases represent reflexive reasoning structure. In this platform, the main control agent organizes agents with various functions and makes them work collaboratively to solve problems through asynchronous communication and cooperation mechanism. Object Oriented paradigm was used to abstract agents that manage resources in the resource base. The computational level is derived from four layers, object layer, software agent layer, intelligent agent layer, and problem solving agent layer. The distributed agent systems offers a transparent control and monitoring of system resources. The architecture is applied to some sample application areas.
The result thus obtained shows that the criticism converged to zero indicating perfect control mechanism. It also indicated that intelligent hybrid multi-agent systems offer more powerful problem solving strategies then conventional control systems.
Testability Models For Structured Programs
Meeta Prakash, A.K. Tripathi
The testability of a software is an important attribute. This work aims at quantification of this attributes of modules of a software system. WE make use of size and complexity attributes of modules of structures software systems for obtaining models that will be useful for development of quantitative measures for the purpose. The necessary description of programs for obtaining intrinsic and directly measurable software component attributes, of modules of structured software systems, are also given. A discussion on relationship between testability and possible mutants of a module has also been included.
Exploitting Parallelism In Gentle Task Allocation Algorithm
Deo Prakash Vidyarthi, Anil Kumar Tripathi
Task allocation in Parallel/Distributed system has become quite challenging problem in the direction of purposeful research in the field. Various algorithms, considering various constraints have been proposed in the past. Genetic Algorithm can be purposefully utilized for solving this problem efficiently. This work aims to identify parallelism present in the Genetic Task Allocation Algorithm. As it is evident, a parallel implementation of the algorithm shall not only reduce the allocation time but also will make use of the platform provided by the Distributed System. We propose the suitable configurations such as pipeline, partitioning etc. It has been demonstrated that various activities in the genetic task allocation algorithm provide inherent parallelism in the same and are amenable to partitioning for configuring corresponding parallel implementation. The various organizations of possible parallel implementations have suitably been depicted.
Cost Effective Indexed Data Broadcasting In Wireless Mobile Computing
Prof. P.C. Saxena, I.J. Arora
Mobile clients are normally more compact and lightweight portable units that are restricted in their energy source. Conservation of battery power is one of the prime issues in wireless mobile computing. Energy can be saved using an indexed data organization to broadcast data over wireless communication channels. In this paper, we consider the index based organization of skewed access data for sequential broadcasting in mobile computing. To minimize the average cost of index probes, we propose a method for variant index fanouts to construct imbalanced index trees according to access frequencies of data. For the proposed method , the decision criteria is computationally less expensive. Also, we explore a factor, which can be used to adjust access frequencies of same size to further reduce the cost. Finally, a condition is observed when an imbalanced index tree takes the form of binary tree.
Assigning Wavelengths in WDM Optical Networks Using An Efficient Graph Coloring Algorithm
Sahadeb Jana and Suman Chowdhury, Mrinal K. Naskar, Debashis Saha, Amitava Mukherjee
The problem of wavelength assignment in WDM optical networks is often posed as a graph-coloring problem. In this paper, we present an efficient algorithm for graph coloring which can be employed for assigning wavelengths in WDM optical networks. Given the physical topology and lightpath requests of a WDM network, we first construct an auxiliary graph, whose nodes represent lightpaths, and then color this auxiliary graph using the above algorithm. The proposed algorithm starts with a temporary color for each node, and, at the termination of the algorithm, each node is assigned a permanent color. We have computed the complexity of the algorithm as O(N 3 ), where N is the number of nodes in the auxiliary graph to be colored.
Nature and The International system : A Mathematical Model of North-South Parity Under Macroclimatic Constraint
Jayantanuja Bandyopadhyaya, Amitava Mukherjee
The international system is assumed to consist of two component units : the North and the South. Nature, in its aspect of the dichotomus global macroclimate, impacts the two units differently from outside the system. While the temperate climate has a positive impact on the economic development of the North, the tropical climate has a negative impact on the economic development of the North, the tropical climate has a negative impact on that of the South. It is postulated that, given the natural constraints, North-South parity can be optimized through a technocultural adaptation on the part of the South involving the replacement of labor-intensive technology by automotive technology, made possible by (1) development of indigenous automotive technology by the South, and (2) transfer of such technology from the North to the South. A mathematical model is developed, initially on the assumption of an anarchical system, of North-South parity as a combinatorial non-linear optimization problem. This problem is solved by using Lagrangean relaxation and subgradient heuristic technique. The model is then extended to include a global regulatory authority.
Analysis of Ordered Dynamic Channel Assignment Scheme with Integrated Traffic
M. Jain and Rakhee
This paper proposes a dynamic channel assignment (DCA) scheme with borrowing for integrated voice/data traffic in cellular radio mobile systems. In dynamic channel assignment schemes, all channels can be made available for every cell in a cluster. It is established that DCA works better than fixed channel assignment (FCA) for non-uniform as well as light traffic. In DCA scheme, the channels can be dynamically assigned to a call in any cell of a cluster subject to co-channel-interference constraint. Due to the large dimension, the exact analysis of DCA is highly complex. In this paper, we suggest ordered-dynamic channel-assignment-with-reassignment (ODCAR) scheme with the provision of guard channels. In this scheme, a group of channels is allocated to each cell of a cluster as in FCA. When all the channels in a particular cell are busy and a hand of call originates, then any channel from whole cluster may be borrowed by the cell, without violating co-channel interference constraint. The expressions for steady state queue size distribution, blocking probabilities for data and voice attempts and the force termination probability etc. are determined. Sensitivity analysis is carried out to examine the effect of varying different parameters on the system performance.
Some Observations on Estimation Metrics for Object-Oriented Software Engineering
A.K. Tripathi, Alok Mishra
Object-Oriented software projects are becoming more popular than structured (functional) technology based projects in the present scenario. Object Technology (OT) offers support to deliver products to market more quickly and provide high quality products with lower maintenance costs. As expertise in Object-Oriented (OO) project grows, such a body of OO metric knowledge will become increasingly usable across the Industry. There is need for good OO metrics for both process and product management. Estimation is an important field of software engineering. In this work we have proposed model for staffing estimation by using available metrics in OO and taking affecting factors into consideration.
|
